IOSR Journal of Mechanical and Civil Engineering (IOSR-JMCE)

e-ISSN: 2278-1684,p-ISSN: 2320-334X, Volume 12, Issue 6 Ver. II (Nov. - Dec. 2015), PP 21-29
www.iosrjournals.org

Process Parameter Optimization of Bead geometry for AISI 446

in GMAW Process Using Grey-based Taguchi Method
U. V. Patil1, Prof. G. R. Naik 2
1

PG. Student, Department of Production Engineering, KITs College of Engineering, Kolhapur

Shivaji University, Kolhapur (India)
2
Asso.Professor
Department of Production Engineering, KITs College of Engineering, Kolhapur
Shivaji University, Kolhapur (India)

Abstract : Gas metal arc welding is process is becoming widespread in the welding industries as it can easily
be applied to variety of metals. Quality of weldment depends upon welding input parameters. This experimental
study aims at developing a multi-objective optimization problem to achieve desired weld bead geometry of
GMAW. Objective functions optimized by Taguchis L9 orthogonal array design and S/N ratio. Welding
voltage, welding speed, wire feed rate and gas flow rate have been selected as the input process parameters.
Quality targets are bead width, bead height and depth of penetration for AISI 446. The Grey relational analysis
in combination with Taguchi approach is used to solve this multi-objective optimization problem. Additionally
the most significant factor identified by using ANOVA.
Keywords -Bead Geometry, Gas Metal Arc Welding, Grey Relational Analysis,Taguchi Orthogonal Array

I.

Introduction

The Gas metal arc welding processes are quickly replacing the shield arc stick electrode process
because the GMAW process can easily be applied to both ferrous and nonferrous metals. GMAW processes are
also popular as they can deposit a large quantity of weld metal in a relatively short period of time. GMAW plays
a significant role in many industries. Welding is mainly important in heavy fabrication industry, which includes
machinery, farming equipment, process tools for manufacturing and mining, and infrastructure for petroleum
refining and distribution. During the welding process, input parameters influenced the weldment quality.
Therefore, welding becomes as a multi-input multi-output process. To catch the preferred weld quality in
GMAW process it is necessary to study input and output process parameters interrelationships. The optimization
and prediction of process parameters is a significant aspect in welding process. This experimental study aims at
developing a multi-objective optimization problem to achieve preferred weld bead geometry for AISI 446 grade
of steel of 8 mm thickness in GMAW process by using grey-based Taguchi method. It is necessary to control
shape of weld bead geometry in determining mechanical properties as it is affected by weld bead geometry
shape. Therefore, proper selection of process parameter is necessary. It is not easy task to developed
mathematical models which gives relationship between input and output process parameters of GMAW process
because there were some unknown nonlinear process parameters [1]. Therefore, it is better to solve this problem
by experimental models. Multiple regression technique was used to found the empirical models for various
welding processes [2, 3]. Datta et at [4] planned multiple regression model and predicted the bead geometry
volume of SAW process. Gunaraj et al [5-6] used five-level factorial techniques to developed mathematical
models for prediction and optimization of weld bead for the SAW process. Kim et al [7] used multiple
regression and neural network to develop an intelligent system for GMAW process. Li et al [8] used SelfAdaptive Offset Network to know the non-linear relationship between the geometry variables and process
parameters of SAW process. Tang et al [9] examined the correlation between process parameters and bead
geometry for TIG welding process using a back propagation neural network. Thao et al [10] using SPSS
window software developed correlation between process parameters and bead geometry parameters for GTAW
process.
.

II. Grey-Based Taguchi Method

For the design of high quality manufacturing system Taguchis philosophy is an effective method. Dr.
Genichi Taguchi has investigated a method based on orthogonal array experiments, which offers much-reduced
variance for the experiment with optimum setting of process control parameters. The combination of design of
experiments with parametric optimization of process to obtain the desired target is achieved in the Taguchi
method. Orthogonal array (OA) offers a set of balanced experimentations and signal-to-noise ratios, which are
the logarithmic functions of desired output serves as objective functions for optimization. This technique aids in
data analysis and prediction of optimum results. Taguchi method uses a statistical measure of performance
DOI: 10.9790/1684-12622129

www.iosrjournals.org

21 | Page

Process Parameter Optimization of Bead geometry for AISI 446 in GMAW Process Using Grey
called signal-to-noise ratio to evaluate optimal parameters settings. The S/N ratio takes both the mean and the
variability into account which is the ratio of the mean to the standard deviation. The quality characteristic of the
process to be optimized defines the ratio. The standard S/N ratio has three criterias, Lower the best (LB),
Nominal is best (NB) and higher the better ((HB). The parameter combination, which has the highest S/N ratio,
gives the optimal setting.
In Grey relational analysis, Grey relational generation is the process which normalizes the experimental
data range from zero to one and it transferring the original data into comparability Sequences. Next, Grey
relational coefficient is calculated based on normalized experimental data to represent the relationship between
the actual and desired experimental data. Overall Grey relational grade be an average of the Grey relational
coefficient of selected responses. Grey relational grade gives the overall performance characteristics of the
multiple response process. This method converts a multiple objective problem into single objective with the
objective function as overall Grey relational grade.

Fig.1Procedure ofgreyrelational analysis.

Grey relational grade is the result of the evaluation of the optimal parametric combination. Taguchi method
gives the optimal factor setting for maximizing overall Grey relational grade. [11]
In Grey relational generation, the normalized bead width, bead height, following to lower-the-better (LB)
criterion can be stated as:
xi (k) =

max ()

(1)

max min ()

Bead penetration should follow larger-the-better criterion, which can be stated as:
Xi(k) =

()min ()

(2)

max ()min ()

Where, xi(k) = the value after the Grey relational generation, min yi(k) = the smallest value of yi(k) for the kth
response, and max yi(k) = the largest value of yi(k) for the kth response. The normalized data after Grey relational
generation are tabularized in table 5. An ideal sequence is x0(k) (k=1, 2, 3......,25) for the responses. Grey
relational grade is to reveal the degree of relation between the 9 sequences [x0(k) and xi(k), i=1, 2, 3........9]
(Table 3).
We can calculate the Grey relational coefficient i(k) as:
i (k)=

min +max

()+max

(3)

where, 0i = x0(k) xi(k) = difference of the absolute value x 0(k) and xi(k),
is the distinguishing coefficient 0 1
min = jminikminx0(k) xj(k)= the smallest value of 0i, and
max= jminikmaxx0(k) xj(k)= largest value of 0i.
The Grey relational grade i can be computed after be an average of the Grey relational coefficients as:
i=

()

DOI: 10.9790/1684-12622129

(4)

www.iosrjournals.org

22 | Page

Process Parameter Optimization of Bead geometry for AISI 446 in GMAW Process Using Grey
where n = number of process responses. The greater value of Grey relational grade relates to extreme
relational degree between the reference sequence x0(k) and the known sequence xi(k). The reference sequence
x0(k) represents the best process sequence; therefore, higher Grey relational grade means that the corresponding
parameter combination is nearer to the optimal. The mean response of the Grey relational grade, its grand mean
and the main effect plot are very significant as optimal process condition may well be evaluated from this plot.

III. Research Methodology

Systematic approach required to develop the mathematical models that predict and control the bead
geometry parameters for GMAW process includes the next steps [12].
3.1 Identification of process parameters plus their limits.
3.2 Development of design matrix
3.3 Conducting the experiments as per design matrix.
3.4 Preparation of specimen
3.5 Measurements of the responses.
3.6 Analysis of experimental data by Grey-based Taguchi method. Following steps are included in analysis.
Determination of Grey relational Generation
Determination of Grey relational coefficient
Determination of Grey relational grade
Determination of response table for Grey relational grade
3.7 S/N ratio plot for overall Grey relational grade
3.8 Confirmation Experiment
3.9 Evaluation of the significance of the factor by analysis of variance (ANOVA)
3.1 Identification of process parameters and their levels in GMAW
The process parameters play very important role to determine the good weld bead. The rough trials are
necessary for a smooth appearance of weld bead and the working ranges of the parameters are found out. If the
functioning ranges are smaller or larger the limits, then proper weld bead will not appear. Therefore, proper
setting with selection of the process parameters is required for good appearance of bead. The four welding input
parameters selected are the welding voltage (V), welding speed (S), wire feed rate (F) and gas flow rate (G) and
the output parameters are bead penetration (BP), bead height (BH) and bead width (BW) as shown in Fig. 2. The
weld bead geometry will influence the mechanical properties of the weldment. These are key controllable
process parameters and it is desirable to have minimum three levels of process parameters to reveal the true
performance of response parameters. The lower and upper limits of parameters are coded as 1 and 3
respectively. Intermediate ranges coded value can be calculated as:
xi = 2(2x-(xmax+ xmin)) / (xmax- xmin)
(5)
where, x i = The required coded value of a parameter x; x is any parameter value from xmax to xmin;
xmin = The lower level of the parameter;
xmax = The upper level of the parameter.
The individual process parameters levels are tabulated in Table 1.

Fig. 2 Input and output process parameters of GMAW

DOI: 10.9790/1684-12622129

www.iosrjournals.org

23 | Page

Process Parameter Optimization of Bead geometry for AISI 446 in GMAW Process Using Grey
Table 1 Process parameters and their limits
Process Parameters
Open Circuit Voltage
Gas Flow Rate
Wire Feed Rate
Welding Speed

Notation
V
G
F
S

Levels of Factors
1
2
26
27
16
17
1.75
2.0
20
22

Unit
Volts
lit/min
cm/min
cm/min

3
28
18
2.25
24

3.2 Development of design matrix

To facilitate the objective of determining the optimal level of each design parameter, Taguchi
popularized the use of orthogonal arrays as an easy way to design fractional factorial experiments. Taguchi
began by selecting several good basic fractional factorial designs and, for each, setting up table, which he
calls an orthogonal array. These tables may be used in very simple ways to design an experiment. For three level
designs, he provides an orthogonal array L9, for up to four factors. Finally, there are methods to use the three
levels tables for 9 runs. The experimental design proceeds are simply the assignment of effects of columns after
the selection of orthogonal array. Table 2 shows the nine experimental runs as per the Taguchis L9 orthogonal
array design matrix. Therefore with four controllable parameters, at three different levels the experimentation
has been carried out.
Table 2 Taguchi L9 Orthogonal array design
Sr. No.

Voltage
(V)

Wire Feed Rate

(cm/min)

Welding Speed
(cm/min)

Gas Flow Rate

(lit/min)

3.3 Conducting the experiments as per design matrix.

In this work, GMAW machine manufactured by ATE, Pune, INDIA is used. It is a 3-phase, 50Hz
frequency, 300A, forced air cooling semi-automatic machine. The torch is fixed to the frame at 900 to the work
specimen can be kept on trolley which will move at perfect path. Gas flow rate can be vary and measured by
flow meter. The mixture of argon (98%) and oxygen gas (2%) is used as a shielding gas. The wire feed and
welding voltage regulators are provided in the machine to changes the values of same. Variation of welding
voltage automatically changes the current. Therefore, four process parameters welding voltage, wire feed rate,
welding speed and gas flow rate can be varied in this machine. Based on design matrix, the experimentations are
done by varying input parameters.
In this present work, AISI 446 grade stainless steel is used as work-piece material. The stainless steel
filler rod of AISI 309L is recommended by American welding Society (AWS) used for conducting the
experiments. The chemical composition of work-piece and filler rod material is shown in Table 3.
Table 3 Chemical Composition of Materials
Material (%)
AISI 446
AISI 309L

C
0.08
0.02

Si
0.35
0.42

Mn
0.85
1.80

S
0.003
0.01

P
0.030
0.02

Cr
23.45
24.80

Ni
0.40
12.92

Mo
.
0.01

Ti

0.124

0.049

Cu

0.03

Nu

0.025

Fe
74.84
59.772

3.4 Preparation of Specimens

To evaluate the bead geometry of the stainless steel weldment, specimens having dimension of 150mm
x 100mm x 8mm with a single V shaped groove of 300 were used for each piece as per the design data on the
100 mm side of the plate using vertical milling machine. Fig 3 shows a schematic diagram of work piece
prepared. The land and root gap of 2mm is provided between the two work pieces. The experiments are
DOI: 10.9790/1684-12622129

www.iosrjournals.org

24 | Page

Process Parameter Optimization of Bead geometry for AISI 446 in GMAW Process Using Grey
conducted using GMAW machine based on Taguchis orthogonal array design matrix which consist of nine
trails.

Fig. 3 Work piece prepared

3.5 Measurements of the response parameters
The experiments are conducted by single pass. The samples are cut into three parts for the
measurement of bead geometry parameters using water jet cutting machine.A schematic diagram of welded
plates is shown in Fig. 4.The samples are prepared as per American Society of Testing Materials (ASTM)
standards. The metallurgical polishing process revealed the bead geometry for measuring the BH, BW and BP.
The silicon carbide of 220, 320 and 500 emery papers and etchant like oxalic acid is used. The bead geometry
parameters are measured at three places and the average values are measured and tabulated in Table 4.

Fig. 4 Welded Samples

Table 4Experimental Data
Sample
No.
1

Bead Width
(mm)
8.14

Bead Height
(mm)
1.44

8.56

2.29

7.66

10.69

3.39

6.42

10.26

3.58

5.94

8.46

2.53

6.27

8.29

3.22

7.06

8.58

2.68

6.48

10.91

2.88

6.42

10.17

3.63

7.43

Bead Penetration (mm)

6.92

3.6 Analysis of experimental data by Grey-based Taguchi method.

The result obtained for bead width, bead height and bead penetration after experimentation have been
first normalized. The bead height and bead width should be minimum and the bead penetration should be
maximum. Therefore lower-the-better (LB) criteria for bead height, bead width using equation 1 and larger-thebetter (LB) criterion for bead penetration using equation 2 has been selected. The normalized data for each of
this parameter are given in Table 5.
DOI: 10.9790/1684-12622129

www.iosrjournals.org

25 | Page

Process Parameter Optimization of Bead geometry for AISI 446 in GMAW Process Using Grey
Table 5 Data preprocessing of all performance characteristic (Grey relational generation)
Experiment No.

Bead Width
(mm)

Bead Height
(mm)

Penetration
(mm)

Ideal Sequence

0.8394

0.9087

0.5000

0.7121

0.5560

0.8776

0.0667

0.0996

0.2449

0.1970

0.0207

0.0000

0.7424

0.4564

0.1684

0.7939

0.1701

0.5714

0.7061

0.3942

0.2755

0.0000

0.3112

0.2449

0.2242

0.0000

0.7602

The 0i for each of the responses have been calculated and tabulated in Table 6. These evaluations of 0i for
each response have been collected to evaluate Grey relational coefficient.
Table 6Evaluation of 0i for each of the responses
Experiment
No.
Ideal Sequence

Bead Width
(mm)
1

Bead Height
(mm)
1

Penetration
(mm)
1

0.1606

0.0913

0.5000

0.2879

0.4440

0.1224

0.9333

0.9004

0.7551

0.8030

0.9793

1.0000

0.2576

0.5436

0.8316

0.2061

0.8299

0.4286

0.2939

0.6058

0.7245

1.0000

0.6888

0.7551

0.7758

1.0000

0.2398

In Table 6, reference sequence is ideal sequence. After calculating max, min and 0i, altogether grey
relational coefficients may be evaluated by (3). These Grey relational coefficients for each response have been
collected to calculate Grey relational grade and are exposed in Table 7.
Table 7Grey relational coefficient of each performance characteristics (with = 0.5)
Experiment No.

Bead Width
(mm)

Bead Height
(mm)

Penetration
(mm)

Ideal Sequence

0.7569

0.8456

0.5000

0.6346

0.5297

0.8033

0.3488

0.3570

0.3984

0.3837

0.3380

0.3333

0.6600

0.4791

0.3755

0.7082

0.3760

0.5385

0.6298

0.4522

0.4083

0.3333

0.4206

0.3984

0.3919

0.3333

0.6759

Grey relational grade is the total characteristic of all the features of weld quality. The overall Grey relational
grade has been determined by (4) and is tabulated in Table 8.
DOI: 10.9790/1684-12622129

www.iosrjournals.org

26 | Page

Process Parameter Optimization of Bead geometry for AISI 446 in GMAW Process Using Grey
Table 8 Grey relational grade
Experiment No.

Grey relational grade

0.8458

0.7768

0.3737

0.3557

0.5652

0.6241

0.5498

0.3916

0.5093

Thus, the multi-objective optimization problem can be transformed into a single comparable objective function
optimization problem by the combination of Taguchi approach and Grey relational analysis. Maximum grey
relational grade equivalents factor combination to the closer optimal parametric setting.
The mean response for the overall Grey relational grade is shown in Table 9.
Table 9Response table (mean) for Overall Grey relational grade
Grey relational grade
Factors
Level 1

Level 2

Level 3

MAX

MIN

Delta

0.6687

0.5300

0.4905

0.6687

0.4905

0.1782

0.5825

0.5941

0.5125

0.5941

0.5125

0.0816

0.6196

0.5602

0.5094

0.6196

0.5094

0.1102

0.6363

0.6763

0.3766

0.6763

0.3766

0.2997

Total mean gray relational grade is 0.5631

Delta = range (maximum-minimum)
Total mean Grey relational grade is the average of all entries in Table 8.
3.7 Calculation of S/N ratio
Fig. 5 represents the S/N ratio for overall Grey relational grade, calculated using LB (larger-the-better) criterion
as:
1
1
SN (Larger-the-better) = 10 log 1 y 2
(6)
Where n = the number of measurements, and yi = the measured characteristic value
The mean response for S/N ratio is tabulated in Table 10.
Table 10Response table for S/N ratio
Factor
V
F
S
G

DOI: 10.9790/1684-12622129

Level 1
-3.4953
-4.6935
-4.1581
-3.9264

Level 2
-5.5146
-4.5223
-5.0324
-3.3974

Level 3
-6.1872
-5.8056
-5.8591
-8.4824

MEAN
-5.0657
-5.0071
-5.0166
-5.2688

www.iosrjournals.org

27 | Page

Process Parameter Optimization of Bead geometry for AISI 446 in GMAW Process Using Grey

M a in E ffe c t s P lo t fo r S N ra t io s
D ata M ean s

-4
-5
-6
-7
-8

M e a n o f S N r a ti o s

26

27

28

1 .7 5

2 .0

2 .2 5

-4
-5
-6
-7
-8
20

22

24

16

17

18

L e ve l s o f F a c t o r s

S ig n a l-t o -n o is e : L a rg e r is b e t t e r

Fig. 5S/N ratio plot for overall Grey relational grade

3.8 Confirmation experiment
After assessing the optimal parameter settings, the next steps are prediction and verify the improvement
of quality targets using the optimal parametric combination. The predictable Grey relational grade using the
optimal level of the design parameters can be determined by table 8 and (7):

=m +

0
=1 (i- m)

(7)

That means the predicted Grey relational grade is equivalent to the mean Grey relational grade plus the
summation of the difference between overall mean Grey relational grade and mean Grey relational grade for
each of the factors at optimal level.
S/N ratio for predictable grey relational grade have been calculated using (6)
S/N ratio for predictable grey relational grade = -1.3313
The Grey relational grade and S/N ratio for Initial factor settings V1 F1 S1 G1, predicted optimal
parametric process setting V1 F1 S1 G2 and experimental optimal parametric process setting V1 F1 S1 G2 are
shown in column no. 2, 3 and 4 respectively in Table 11.
Table 11 signifies the comparison of the predicted bead geometry parameters with that of actual by
using the optimal welding conditions; improvement in overall Grey relational grade has been observed. This
confirms the usefulness of the proposed approach in relation to process optimization, where more than one
objective has to be fulfilled concurrently.The overall Grey relational grade is main performance feature in Greybased Taguchi method; The Grey relational grade is the demonstrative of all individual performance
characteristics. In the current study, objective functions selected are relative to parameters of bead geometry;
and all the responses give equal weight age. It may be noted that results of a confirmatory experiment shows
that using optimal parameter setting the weldment undertakes lower value of depth of penetration. The Taguchi
optimization technique and the optimal parametric combination thus calculated depend on the selected response
variables and their individual weight ages.
Table 11 Results of confirmatory experiment
Initial factor setting
Levels of factors
Bead Width
Bead Height
Penetration
S/N ratio of overall Grey relational grade
Overall Grey relational grade
Improvement in grey relational grade = 0.0158

DOI: 10.9790/1684-12622129

V1 F1 S1 G1
8.14
1.44
6.92
-1.7460
0.8179

Optimal process condition

Prediction
Experiment
V1 F1 S1 G2
V1 F1 S1 G2
9.26
1.2
7.99
-1.3313
-1.5798
0.8579
0.8337

www.iosrjournals.org

28 | Page

Process Parameter Optimization of Bead geometry for AISI 446 in GMAW Process Using Grey
3.9 Analysis of variance (ANOVA)
ANOVA is a statistical technique, which can infer some important conclusions based on analysis of the
experimental data. If the P-value for a term appears less than 0.05 then it may be concluded that, the effect of
the factor is significant on the selected response. Using MINITAB release 15, ANOVA for overall Grey
relational grade has been exposed in Table 12. It is saw that P-value for Gas flow rate is 0.040 (less than 0.05).
So it is obvious that gas flow rate is the most significant factor. The effects of other factors on Grey relational
grade look insignificant. Gas flow rate G (P-value highest) come to be the most insignificant factor.
Table 12 Analysis of variance using adjusted SS for tests
Source
V
F
S
G
Error

DF
2
2
2
2
6

SS
0.0526
0.0117
0.0182
0.1589
0.1888

Total

14

0.2414

MS
0.0263
0.0058
0.0091
0.0794
0.315

F
0.84
0.15
0.25
5.78

P
0.479
0.862
0.790
0.040

IV. Conclusion
The following conclusions are from the experimentation:
1. Grey based Taguchi method was successfully applied for process parameter optimization of bead geometry
for AISI 446 in GMAW process.
2. In order to achieve the best bead geometry viz. maximum bead penetration and minimum bead width as
well as bead height following optimal settings were determined: Welding Voltage 26 V, wire feed rate1.75 cm/min, welding speed- 20 cm/min and gas flow rate- lit/min.
3. Gas flow rate was identified as the most significant factor for bead geometry using ANOVA.

References
[1].
[2].
[3].
[4].
[5].
[6].
[7].
[8].
[9].
[10].
[11].
[12].

D. Kim, M. Kang, S. Rhee, Determination of optimal welding conditions with a controlled random search procedure, Welding
Journal 8 (2005) 125-130.
J. Raveendra, R.S. Parmar, Mathematical models to predict weld bead geometry for flux cored arc welding, Metal Construction 19/2
(1987) 31-35.
L.J. Yang, R.S. Chandel, M.J. Bibby, The effects of process variables on the weld deposit area of submerged arc welds, Welding
Journal 72/1 (1993) 11-18.
S. Datta, M. Sundar, A. Bandyopadhyay, P.K. Pal, S.C. Roy, G. Nandi, Statistical modeling for predicting bead volume of
submerged arc butt welds, Aust. Welding Journal 51 (2006) 4-8.
V. Gunaraj, N. Murugan, Prediction and optimization of weld bead volume for the Submerged Arc Process - Part 1, Welding
Journal 10 (2000) 286-294.
V. Gunaraj, N. Murugan, Prediction and optimization of weld bead volume for the Submerged Arc Process - Part 2, Welding
Journal 11 (2000) 331-338.
I.S. Kim, J.S. Son, C.E. Park, I.J. Kim, H.H. Kim, An investigation into an intelligent system for predicting bead geometry i n GMA
welding process, Journal of Materials Processing Technology 159/1 (2005) 113-118.
P. Li, M.T.C. Fang, J. Lucas, Modelling of submerged arc welding bead using self-adaptive offset neural network, Journal of
Materials Processing Technology 71 (1997) 228-98.
Y.S. Tang, H.L. Tsai, S.S. Yeh, Modelling, optimization and classification of weld quality in tungsten inert gas welding, Journal of
Machine Tools Manufacture 39 (1999) 1427-38.
D.T. Thao,J.W. Jeong, I.S. Kim, J.W.H.J. Kim, Predicting Lap-Joint bead geometry in GMA welding process,International
Scientific Journal,32(2008)121-124.
Saurav Datta, Ashish Bandyopadhyay, Pradeep Kumar Pal. Grey based Taguchi method foroptimization of bead geometry in
submerged arc bead-on-plate welding.International journal of Advance Manufacturing Technology (2008) 39:1136-1143.
P.K.Palani, N.Murugan, Optimization of weld bead geometry for stainless steel claddings deposited by FCAW, Journal of Material
Processing Technology 190(2007) 291-299.

DOI: 10.9790/1684-12622129

www.iosrjournals.org

29 | Page